Switch back to ConcurrentMergeScheduler

Load tests showed that SerialMS has problems to keep up with
the merges under high load. We should switch back to CMS
until we have a better story to balance merge
threads / efforts across shards on a single node.

Closes #5817

Author: s1monw

docs/reference/index-modules/merge.asciidoc

@@ -187,11 +187,7 @@ Defaults to unbounded.
 
 The merge schedule controls the execution of merge operations once they
 are needed (according to the merge policy). The following types are
-supported, with the default being the `SerialMergeScheduler`.
-
-Note, the default is the serial merge scheduler since there is a merge
-thread pool that explicitly schedules merges, and it makes sure that
-merges are serial within a shard, yet concurrent across multiple shards.
+supported, with the default being the `ConcurrentMergeScheduler`.
 
 [float]
 ==== ConcurrentMergeScheduler

src/main/java/org/elasticsearch/index/merge/scheduler/MergeSchedulerModule.java

@@ -28,7 +28,7 @@
 public class MergeSchedulerModule extends AbstractModule {
 
     public static final String MERGE_SCHEDULER_TYPE_KEY = "index.merge.scheduler.type";
-    public static final Class<? extends MergeSchedulerProvider> DEFAULT = SerialMergeSchedulerProvider.class;
+    public static final Class<? extends MergeSchedulerProvider> DEFAULT = ConcurrentMergeSchedulerProvider.class;
 
     private final Settings settings;